A dynamic compensation method for machining accuracy of turning and milling machine tools

Abstract

The present invention relates to the field of machine tool processing technology, and specifically to a method for dynamic compensation of processing accuracy of a turning-milling compound machine tool. An error prediction LSTM model is used to perform deep learning on time series data. The system can identify the potential patterns and laws in the processing process, and is suitable for processing complex data with time series. The system can effectively analyze the correlation between temperature, vibration and displacement, and convert it into accurate error prediction. The cutting parameters are dynamically adjusted through the error prediction value to achieve the effect of dynamic compensation of processing accuracy.

Description

Technical Field

[0001] The invention relates to the technical field of machine tool processing, and more specifically, to a dynamic compensation method for processing accuracy of a turning and milling compound machine tool. Background Art

[0002] With the continuous development of the mechanical processing industry and the increasing requirements for product quality, processing accuracy has become an important factor affecting the competitiveness of enterprises. During the processing of traditional turning and milling compound machine tools, due to the influence of various dynamic factors such as temperature, vibration and tool displacement, the processing error often increases, thereby affecting the accuracy and surface quality of parts. The existing methods for adjusting the accuracy, such as a CNC machine tool thermal deformation test compensation method with patent publication number CN113297761B, only adjust the compensation according to the processing temperature. The compensation adjustment accuracy is limited and it is difficult to meet the requirements of higher precision processing. Therefore, how to deeply analyze and dynamically compensate for the errors in the machine tool processing process has become a problem to be solved in the current mechanical processing field.

[0003] To this end, we propose a dynamic compensation method for the machining accuracy of turning-milling machine tools, which monitors and provides feedback on the machine tool status in real time. Through sensors, we collect real-time data including temperature, vibration, tool displacement and other dynamic factors, and dynamically adjust the machine tool operation status to achieve intelligent prediction of machining errors and timely adjust machine tool parameters, thereby achieving dynamic optimization of machine tool machining accuracy. Summary of the invention

[0004] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a dynamic compensation method for machining accuracy of a turning-milling compound machine tool to solve the problems existing in the above-mentioned background technology.

[0005] The above technical objectives of the present invention are achieved through the following technical solutions: A dynamic compensation method for machining accuracy of a turning-milling compound machine tool, comprising:

[0006] S1. A sensor detection module is provided on the machine tool to detect and obtain the working parameters of the machine tool; the working parameters include the working temperature of the tool side of the machine tool, the vibration coefficient of the machine tool and the displacement coefficient of the tool cutting process;

[0007] S2. Construct a data set using the working parameter values ​​obtained in step S1;

[0008] S3, building an error prediction LSTM model using the data set constructed in step S2;

[0009] S4, predicting the processing error at the next processing time point through the constructed error prediction LSTM model;

[0010] S5. Compare the processing error predicted by the error prediction LSTM model with the target error value of the processing to obtain the target deviation value of the processing;

[0011] S6, the analysis module dynamically generates a machine tool cutting parameter adjustment instruction through a PID algorithm according to the target deviation value calculated in step S5;

[0012] S7, the execution module receives the machine tool cutting parameter adjustment instruction generated in step S6 and adjusts the machine tool cutting parameters to achieve dynamic adjustment of the machine tool to improve the machining accuracy of the machine tool.

[0013] Optionally, the sensing detection module includes: a non-contact detection method for detecting the working temperature of the machine tool; A temperature sensor for detecting the vibration coefficient of the machine tool A vibration sensor and a displacement coefficient for detecting the tool side A displacement sensor;

[0014] Said is an integer not less than 3; and are integers not less than 2.

[0015] Optionally, the working parameter values ​​obtained in step S1 in step S2 are used to construct a data set, and the specific implementation process is as follows:

[0016] 2.1. Define the data set D according to the working parameter values ​​of the recording point i in the acquisition cycle. ; Where n is the number of all recording points i in the acquisition period;

[0017] 2.2、Data points are ,in, Represented as the first record point The temperature data collected by the temperature sensor is Represented as the first record point The vibration data collected by the vibration sensor, Represented as the first record point The displacement data collected by the displacement sensor.

[0018] Optionally, in step S3, the data set constructed by step S2 is used to establish an error prediction LSTM model, and the specific implementation process is as follows:

[0019] 3.1. Standardize the imported data sets, calculate the mean and standard deviation of each data set, and convert each data set into feature data with a mean of 0 and a standard value of 1;

[0020] 3.2. Generate time series samples from the standardized feature data, and use the generated time series samples as input samples;

[0021] 3.3. Define the shape of the input sample. The shape of the sample is (number of samples, number of time steps, number of features).

[0022] 3.4. Build an LSTM model: Set up a neural network architecture with multiple LSTM layers, add a fully connected layer at the end, set the output dimension to 1 to adapt to the regression problem, and build an error prediction LSTM model;

[0023] 3.5. Train the LSTM model with several input samples to obtain an error prediction LSTM model.

[0024] Optionally, the step S4 predicts the processing error at the next processing time point by using the constructed error prediction LSTM model, and the specific implementation process is as follows:

[0025] 4.1. Define the time step T, which is the data of T time steps before the prediction time point;

[0026] 4.2. Input the data of T time steps before the prediction time point as input samples into the error prediction LSTM model to obtain the target sample Y, which is the target value at the prediction time point;

[0027] 4.3. Calculate the target deviation value through the target sample Y .

[0028] Optionally, the processing error predicted by the error prediction LSTM model in step S5 is compared with the target error value of the processing to obtain the target deviation value of the processing. The specific implementation process is as follows:

[0029] 5.1. Calculate the difference between the target deviation value and the processing error value. ,in is the target deviation value, is the set processing error value;

[0030] 5.2 When If ≤0, it is judged that no adjustment is required;

[0031] 5.3 When >0, it is judged that adjustment is needed and the target deviation value is Output.

[0032] Optionally, the analysis module in step S6 dynamically generates a machine tool cutting parameter adjustment instruction through a PID algorithm based on the target deviation value calculated in step S5. The specific implementation process is as follows:

[0033] 6.1、The target deviation value Input into the PID algorithm to obtain the control output value ,in Defined as the proportional gain, Defined as the integral gain, it is used to eliminate the steady-state error, Defined as the differential gain, predicting future error changes, Defined as the cumulative value of the error value up to the current time point;

[0034] 6.2. The analysis module will control the output value Convert into cutting parameter adjustment instructions for machine tools and output;

[0035] 6.3. Control output value Convert to over-cutting adjustment command: ,in is an adjustable parameter;

[0036] 6.4. Control output value Convert to feed depth adjustment command: ,in is the proportional coefficient of the machine tool tool depth change.

[0037] In summary, the present invention has the following beneficial effects:

[0038] 1. The method of the present invention uses the error prediction LSTM model to perform deep learning on time series data. The system can identify the hidden patterns and laws in the processing process, is suitable for processing complex data with time series, and can effectively analyze the correlation between temperature, vibration and displacement, and convert it into accurate error prediction.

[0039] 2. The method of the present invention implements the PID (proportional-integral-differential) control strategy, which can dynamically generate cutting parameter adjustment instructions in a very short time according to the target deviation value fed back in real time, so that the machine tool always maintains the optimal working state; the dynamic compensation mechanism ensures the processing stability of the machine tool during continuous operation. The real-time feedback and adjustment of the system make the processing process of the machine tool more reliable, and can maintain a small error fluctuation in long-term production. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a flow chart of the method steps of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings. Several embodiments of the present invention are shown in the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein.

[0042] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances. The terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0043] In the present invention, unless otherwise clearly specified and limited, the first feature being "above" or "below" the second feature may include the first feature and the second feature being in direct contact, or may include the first and second features not being in direct contact but being in contact through another feature between them. Moreover, the first feature being "above", "above" and "above" the second feature includes the first feature being directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. The first feature being "below", "below" and "below" the second feature includes the first feature being directly below and obliquely below the second feature, or simply indicates that the first feature is lower in level than the second feature. The terms "vertical", "horizontal", "left", "right", "above", "below" and similar expressions are for illustrative purposes only, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention.

[0044] The present invention is described in detail below in conjunction with the accompanying drawings and embodiments.

[0045] The present invention provides a dynamic compensation method for machining accuracy of a turning and milling compound machine tool, such as Figure 1 As shown, including:

[0046] S1. A sensor detection module is provided on the machine tool to detect and obtain the working parameters of the machine tool; the working parameters include the working temperature of the tool side of the machine tool, the vibration coefficient of the machine tool and the displacement coefficient of the tool cutting process;

[0047] S2. Construct a data set using the working parameter values ​​obtained in step S1;

[0048] S3, building an error prediction LSTM model using the data set constructed in step S2;

[0049] S4, predicting the processing error at the next processing time point through the constructed error prediction LSTM model;

[0050] S5. Compare the processing error predicted by the error prediction LSTM model with the target error value of the processing to obtain the target deviation value of the processing;

[0051] S6, the analysis module dynamically generates a machine tool cutting parameter adjustment instruction through a PID algorithm according to the target deviation value calculated in step S5;

[0052] S7, the execution module receives the machine tool cutting parameter adjustment instruction generated in step S6 and adjusts the machine tool cutting parameters to achieve dynamic adjustment of the machine tool to improve the machining accuracy of the machine tool.

[0053] In the specific implementation process, the analysis module includes an error calculation module and a PID controller, the target deviation value is received and calculated by the error calculation module and input into the PID controller, and based on the calculated target deviation, a cutting parameter adjustment instruction is generated using a PID algorithm;

[0054] The execution module includes an instruction parsing module, which is used to receive and parse the cutting parameter adjustment instructions transmitted by the analysis module; a machine tool control unit, which is responsible for converting the parsed instructions into machine tool operations and adjusting the speed, feed rate, and cutting depth; a feedback monitoring unit, which monitors the cutting process of the machine tool in real time to ensure that it maintains the set cutting parameters during operation and feeds back the cutting data to the analysis module in real time.

[0055] Furthermore, the sensing detection module includes: a non-contact detection method for detecting the working temperature of the machine tool; A temperature sensor for detecting the vibration coefficient of the machine tool A vibration sensor and a displacement coefficient for detecting the tool side A displacement sensor;

[0056] Said is an integer not less than 3; and are integers not less than 2.

[0057] In the specific implementation process, when there are three temperature sensors, the installation positions are: installed near the surface of the machine tool cutting tool to monitor the heat during the cutting process; installed on the spindle body of the machine tool to monitor the temperature changes of the spindle motor and bearings of the machine tool; installed on the cutting fluid box to monitor the temperature of the cutting fluid to ensure that it is within the normal working range; the remaining temperature sensors can be installed in the precise parts where the temperature needs to be monitored to obtain better temperature data;

[0058] When there are two vibration sensors, the installation positions are: Installed on the machine tool spindle to monitor the operating status of the spindle and determine the wear and imbalance; Installed on the bottom bracket to measure the overall vibration of the machine tool to determine the degree of interference of external factors on the operation of the machine tool; The remaining vibration sensors can be installed in the precise parts where the vibration coefficient needs to be monitored to obtain better vibration data and ensure the influence of vibration;

[0059] When there are two tool displacement sensors, the installation positions are: installed at the front end of the machine tool tool to monitor the feed displacement of the machine tool tool to ensure cutting accuracy; installed on the side of the machine tool tool to monitor the lateral displacement of the machine tool tool to avoid overcutting.

[0060] Furthermore, the working parameter values ​​obtained in step S1 in step S2 are used to construct a data set, and the specific implementation process is as follows:

[0061] 2.1. Define the data set D according to the working parameter values ​​of the recording point i in the acquisition cycle. ; Where n is the number of all recording points i in the acquisition period;

[0062] 2.2、Data points are ,in, Represented as the first record point The temperature data collected by the temperature sensor is Represented as the first record point The vibration data collected by the vibration sensor, Represented as the first record point The displacement data collected by the displacement sensor.

[0063] In the specific implementation process, if the number of temperature sensors is set , the number of vibration sensors , the number of tool displacement sensors , the data collection period is set to 60s, the sampling interval is 1s, then n=60;

[0064] The representation of the dataset D is , each data point The expression of .

[0065] Furthermore, in step S3, the data set constructed in step S2 is used to establish an error prediction LSTM model, and the specific implementation process is as follows:

[0066] 3.1. Standardize the imported data sets, calculate the mean and standard deviation of each data set, and convert each data set into feature data with a mean of 0 and a standard value of 1;

[0067] 3.2. Generate time series samples from the standardized feature data, and use the generated time series samples as input samples;

[0068] 3.3. Define the shape of the input sample. The shape of the sample is (number of samples, number of time steps, number of features).

[0069] 3.4. Build an LSTM model: Set up a neural network architecture with multiple LSTM layers, add a fully connected layer at the end, set the output dimension to 1 to adapt to the regression problem, and build an error prediction LSTM model;

[0070] 3.5. Train the LSTM model with several input samples to obtain an error prediction LSTM model.

[0071] In the specific implementation process, the imported data set is standardized. ,in is the mean of each data set, Represented as the standard deviation of each data set, each data set is converted into feature data with a mean of 0 and a standard value of 1 ;

[0072] At the same time, create an input sequence, that is, create a time window for each input feature data. , plan the input feature data to facilitate subsequent training;

[0073] The established LSTM model is constructed through multiple LSTM layers and fully connected layers, defining input and output. The input is , the output is , where W represents the weight of the fully connected layer and b is the bias;

[0074] When training the LTSM model, in order to ensure the accuracy of the data and reduce errors, the LTSM model is optimized, and the optimization goal is to minimize the loss function. ,in Represents the actual error value of the i-th training sample, It is expressed as the prediction error value of the i-th training sample to obtain the error prediction LSTM model.

[0075] Furthermore, the error prediction LSTM model constructed in step S4 predicts the processing error at the next processing time point, and the specific implementation process is as follows:

[0076] 4.1. Define the time step T, which is the data of T time steps before the prediction time point;

[0077] 4.2. Input the data of T time steps before the prediction time point as input samples into the error prediction LSTM model to obtain the target sample Y, which is the target value at the prediction time point;

[0078] 4.3. Calculate the target deviation value through the target sample Y .

[0079] In the specific implementation process, after defining the time step T, input the input feature data that matches the time step T , by inputting feature data into the error prediction LSTM model, the error prediction of the next processing time point is predicted. .

[0080] Furthermore, the processing error predicted by the error prediction LSTM model in step S5 is compared with the target error value of the processing to obtain the target deviation value of the processing. The specific implementation process is as follows:

[0081] 5.1. Calculate the difference between the target deviation value and the processing error value. ,in is the target deviation value, is the set processing error value;

[0082] 5.2 When If ≤0, it is judged that no adjustment is required;

[0083] 5.3 When >0, it is judged that adjustment is needed and the target deviation value is Output.

[0084] In the specific implementation process, The processing error value allowed for this processing operation can be set and adjusted manually. The error value predicted by the error prediction LTSM model is obtained by comparing the two. ,pass to determine whether adjustments are needed.

[0085] Furthermore, the analysis module in step S6 dynamically generates a machine tool cutting parameter adjustment instruction through a PID algorithm according to the target deviation value calculated in step S5. The specific implementation process is as follows:

[0086] 6.1、The target deviation value Input into the PID algorithm to obtain the control output value ,in Defined as the proportional gain, Defined as the integral gain, it is used to eliminate the steady-state error, Defined as the differential gain, predicting future error changes, Defined as the cumulative value of the error value up to the current time point;

[0087] 6.2. The analysis module will control the output value Convert into cutting parameter adjustment instructions for machine tools and output;

[0088] 6.3. Control output value Convert to over-cutting adjustment command: ,in is an adjustable parameter;

[0089] 6.4. Control output value Convert to feed depth adjustment command: ,in is the proportional coefficient of the machine tool tool depth change.

[0090] In the specific implementation process, the execution module converts and The cutting parameters of the machine tool are controlled and adjusted to achieve dynamic compensation of the machining accuracy to achieve higher precision.

[0091] In the above embodiment, further implementation process, by predicting The function is adjusted with the actual error generated at the processing time point to optimize the error prediction LTSM model, thereby further improving and optimizing the LTSM model.

[0092] The present invention discloses a method for dynamically compensating machining accuracy of a turning-milling compound machine tool. The method of the present invention utilizes an error prediction LSTM model to perform deep learning on time series data. The system can identify the patterns and laws hidden in the machining process and is suitable for processing complex data with time series. The system can effectively analyze the correlation between temperature, vibration and displacement and convert it into accurate error prediction. The cutting parameters are dynamically adjusted through the error prediction values ​​to achieve the effect of dynamically compensating machining accuracy.

[0093] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A dynamic compensation method for machining accuracy of a turning-milling compound machine tool, characterized in that: include: S1. A sensor detection module is provided on the machine tool to detect and obtain the working parameters of the machine tool; the working parameters include the working temperature of the tool side of the machine tool, the vibration coefficient of the machine tool and the displacement coefficient of the tool cutting process; S2. Construct a data set using the working parameter values ​​obtained in step S1; S3, building an error prediction LSTM model using the data set constructed in step S2; S4, predicting the processing error at the next processing time point through the constructed error prediction LSTM model; S5. Compare the processing error predicted by the error prediction LSTM model with the target error value of the processing to obtain the target deviation value of the processing; S6, the analysis module dynamically generates a machine tool cutting parameter adjustment instruction through a PID algorithm according to the target deviation value calculated in step S5; S7, the execution module receives the machine tool cutting parameter adjustment instruction generated in step S6 and adjusts the machine tool cutting parameters to achieve dynamic adjustment of the machine tool to improve the machining accuracy of the machine tool; The error prediction LSTM model constructed in step S4 predicts the processing error at the next processing time point, and the specific implementation process is as follows: 4.

1. Define the time step T, which is the data of T time steps before the prediction time point; 4.

2. Input the data of T time steps before the prediction time point as input samples into the error prediction LSTM model to obtain the target sample Y, which is the target value at the prediction time point; 4.

3. Calculate the target deviation value through the target sample Y ; The processing error predicted by the error prediction LSTM model in step S5 is compared with the target error value of the processing to obtain the target deviation value of the processing. The specific implementation process is as follows: 5.

1. Calculate the difference between the target deviation value and the processing error value. ,in is the target deviation value, is the set processing error value; 5.2 When If ≤0, it is judged that no adjustment is required; 5.3 When >0, it is judged that adjustment is needed and the target deviation value is Output; The analysis module in step S6 uses the target deviation value calculated in step S5 to dynamically generate a machine tool cutting parameter adjustment instruction through a PID algorithm. The specific implementation process is as follows: 6.1、The target deviation value Input into the PID algorithm to obtain the control output value ,in Defined as the proportional gain, Defined as the integral gain, it is used to eliminate the steady-state error, Defined as the differential gain, predicting future error changes, Defined as the cumulative value of the error value up to the current time point; 6.

2. The analysis module will control the output value Convert into cutting parameter adjustment instructions for machine tools and output; 6.

3. Control output value Convert to over-cutting adjustment command: ,in is an adjustable parameter; 6.

4. Control output value Convert to feed depth adjustment command: ,in is the proportional coefficient of the machine tool tool depth change.

2. The method for dynamic compensation of machining accuracy of a turning-milling compound machine tool according to claim 1, characterized in that: The sensing detection module includes: a non-contact detection method for detecting the working temperature of the machine tool A temperature sensor for detecting the vibration coefficient of the machine tool A vibration sensor and a displacement coefficient for detecting the tool side A displacement sensor; Said are all integers not less than 3; and are integers not less than 2.

3. The method for dynamic compensation of machining accuracy of a turning-milling compound machine tool according to claim 1, characterized in that: The step S2 uses the working parameter values ​​obtained in step S1 to construct a data set, and the specific implementation process is as follows: 2.

1. Define the data set D according to the working parameter values ​​of the recording point i in the acquisition cycle. ; Where n is the number of all recording points i in the acquisition period; 2.2、Data points are ,in, Represented as the first record point The temperature data collected by the temperature sensor Represented as the first record point The vibration data collected by the vibration sensor, Represented as the first record point The displacement data collected by the displacement sensor.

4. The method for dynamic compensation of machining accuracy of a turning-milling compound machine tool according to claim 1, characterized in that: In step S3, the data set constructed by step S2 is used to establish an error prediction LSTM model, and the specific implementation process is as follows: 3.

1. Standardize the imported data sets, calculate the mean and standard deviation of each data set, and convert each data set into feature data with a mean of 0 and a standard value of 1; 3.

2. Generate time series samples from the standardized feature data, and use the generated time series samples as input samples; 3.

3. Define the shape of the input sample. The shape of the sample is the number of samples, the number of time steps, and the number of features. 3.

4. Build an LSTM model: Set up a neural network architecture with multiple LSTM layers, add a fully connected layer at the end, set the output dimension to 1 to adapt to the regression problem, and build an error prediction LSTM model; 3.

5. Train the LSTM model with several input samples to obtain an error prediction LSTM model.

Images